benchmarks

AI

OpenAI's Astra Model Tests AI Cyber Capabilities

OpenAI will roll out Astra, its first model with critical cyber abilities, sparking security debates as rivals push privacy‑first and benchmark‑driven AI.

AI

GitHub to charge Copilot by usage as inference costs rise

GitHub moves to usage‑based billing for Copilot, Anthropic backs Blender, and Snorkel funds open AI benchmarks, highlighting shifting economics in the AI stack.